American Mineralogist, Volume 73, pages 140-144, 1988
Optical spectrum, site occupancy, and oxidation state of Mn in montmorillonite
Davrn M. Snnnvr,qw.Nonua Vnnco
U.S. Geological Survey, 959 National Center, Reston, Virginia 22092,U.5.A.
AssrRAcr
Optical (difftrse-reflectance)
spectraare used to characterizethe crystal chemistry of Mn
in pegmatite-associatedmontmorillonite. The optical spectra indicate that Mn is in the
3* oxidation stateand show Mn3+ ligand-fieldtransitionsnear 10400, 18800, 20600,
and22400 cm-'. As indicated from the number and eneryiesof the ligand-field absorption
bands, the Mn3* coordination site has undergonea tetragonal distortion associatedwith
the Jahn-Teller effect. Further distortion has eliminated even the tetragonal symmetry.
Assuming that the Mn site is approximately tetragonal, however, the ligand-field theory
parametersloDq, Ds, and Dt are estimatedto be 18800, 1934,and 533 cm '. The large
Jahn-Teller distortion of the Mn3+ coordination site may explain the limited solubility of
Mn3+ cations in montmorillonite and the absenceof a manganian (Mn3+) smectite endmember phase.
INrnouucrroN
Montmorillonites found as pocket clay in granitic pegmatites often have relatively high Mn contents (>0. lolo
as MnO). Such pegmatiteclayshave a characteristicrosepink color that has been attributed to the presenceof Mn
(e.g.,Foord et al., 1986).It is generallyassumedthat Mn
in smectitesis in the Mnr* oxidation state (e.g.,Weaver
and Pollard, 1973). However, higher oxidation statesof
Mn are often found in other pegmatite minerals (e.g.,
Mn3+, Mn++ in tourmaline) and the possibitityof Mn3+
or Mna+ cations in montmorillonite should be considered. Knowing the redox state of Mn in pegmatite clay
minerals may help constrain the oxygen fugacity during
cooling and subsequentalteration of primary pegmatite
minerals.
Optical spectraprovide a useful means of characteizing the crystal chemistry of transition-metal cations in
minerals. For clay minerals, which do not form crystals
large enough for measuring polarized single-crystal absorption spectra,the most useful technique is difuse-reflectancespectroscopy.In this paper, visible to near-infrared diffuse-reflectancespectra are use to characterize
the crystal chemistry of Mn in pegmatite clay.
ExpBnrilmNTAL DETAILS
Materialsandmethods
Two of thesamples
analyzed
in this studyarefrom the SmithsonianInstitution(NationalMuseumof Natural History)collection.NMNH 101836is from Veracruz,Mexico,andNMNH
R7452is from Greenwood,Maine.Thesesampleswereoriginally describedby Rossand Hendricks(1945).A third sample
(K-P)is from the Katrinamine(Paladistrict),SanDiegoCounty, California,andwasdescribed
by Foordet al. (1986).In each
of theselocalities,the montmorilloniteoccupiesfracturesand
pocketsin pegmatitebodies.
Detailedcharacterization
of the sampleswasdoneto confirm
0003-004x/8
8/0I 024 I 40$02.00
that the Mn was not associatedwith some minor accessoryphase.
Samples were dispersed in distilled water using an ultrasonic
probe and size separatedby centrifugation to yield ttre <2-pm
size fraction. Difractometer scansof random-orientation mounts
of the <2-pm size fraction were done using Ni-filtered Cu radiation. This size fraction, which contained only smectite, was
then oriented by centrifugation on a ceramic tile, following the
methods of Kinter and Diamond (1956). X-ray diffraction patterns were run on untreated,glycol-saturated,K-saturated, Mgsaturated,and heat-treated(300'C for 2 h and 550 "C for 2 h
samples.Sampleswere saturatedwith Mg and K using the proceduresoutlined in Jackson(1956).Glycol solvationwasachieved
by direct application ofethylene glycol to the clay on the ceramic
tile.
Chemical compositions were obtained using X-ray fluorescence.Ball-milled sampleswere mixed with the lithium metaborate in a l:10 ratio, fired at 900'C, and molded into a 20mm-wide disc. Oxide abundanceswere determined by a ZAF
algorithm. Total HrO and CO, contents were measuredusing a
CHN analyzer.
Visible to near-infrared (350-2600 mm) diffuse-reflectance
spectra were obtained using a Beckman UV5240 spectrophotometer with a Halon-coated integrating sphere.The spectrawere
first convertedto a Kubulka-Munk remission function, which is
defined as,F(R) : (1 - R),/zR = k/s, where R is reflectance,k
is absorption coefficient, and s is the scatteringcoefrcient. To
the extent that the scatteringcoefficientis independentofwavelength, the Kubulka-Munk remission function will approximate
the absorption spectrum. Each converted spectrum was fit to a
sum of Gaussianbands on a constant baseline.The fitting parameterswere entirely unconstrained.
Mineralogical
and chemical results
The two Smithsonian samples consist predominantly
ofdioctahedral smectite based on (060) d spacings of 1.491.50 A fiable l). NMNH R7452 conrains po[ucite (a
Cs-rich zeolite); NMNH 101836 contains only quartz as
an accessoryphase. Sample K-P (Foord et a1., 1986) contains about 200/oadmixed cookeite.
140
WAVELENGTH (NM)
WAVELENGTH (NM)
NMNH 101436
6
VERACBUZ,
v
1000
500
1000
500
r4l
AND VERGO: Mn IN MONTMORILLONITE
SHERMAN
6
5
MEX.
u,
z
trl
(J
z
o
tr
o
a
o
(D
c
o
a
o
15
10
15
10
W A V E N U M B El O
RO
S /O
W A V E N U M B El O
RO
S /O
Fig. l. Diffirse-reflectance
spectraand modeldeconvolution Fig. 2. Difuse-reflectance
spectraof NMNH R7452.Bands
Bandsat10400,18800,20600,
and2240O in the spectraareasin Fig. l.
ofNMNH 101836.
transitionsof Mn3+. Theweak,sharp
cm-r resultfrom electronic
bandsnear9000cm I are overtoneand combinationbandsof
sharp absorption features interfere with the broad band
interlayerHrO and structuralOH.
centerednear 10400 cm-'. These featuresare due to
overtone and combination vibrational modes of interlayAll of the sampleshave an interlayer thicknessof about er HrO and structural OH.
The energies,widths, and relative intensities of the
two water layers (doo,t tS A) and expand along c* to
glycol
found in the montmorillonite spectra are characK
bands
saturated. Upon
saturation,
about 17 A when
rhe (001) spacingcollapsesto 12.8 A NilvlNH R7452) teristic of Mn3* but not Mn2+. A similar set of absorption
and 13.8 A NvtNH 101836)indicating a low negative bandsis found in manganian(Mn3+ andalusite(Halenius,
charge on the silicate lattice characteristic of smectite 1978)and in piemontite (Burns, 1970).In contrast,the
rather than vermiculite. Heat treatment producesfurther three lowest-energy absorption bands observed in the
collapseto 10.16and 9.93 A. A normal smectiteheated spectraof Mn2+ silicatesare found near I 9 400 cm-' (5 I 5
to 550'C for 2 h should collapseto a doo,of l0 A or less n m ) , 2 3 0 0 0 c m - ' ( 4 3 5 n m ) , a n d 2 4 5 0 0c m ' ( 4 0 8 n m )
(e.g.,Keesterand White, 1968).Those bandsare due to
(Brindley, 1980).A doo,spacingof 10.16 A m NvfNH
I 0 I 836 suggeststhe presenceof minor interlayer hydrox- the 6A, - oTr,6A, ' aTz, lfLd uA, ' oE,uArligand-field
ycatlon.
transitions of octahedrally coordinated Mn2+. No eviStructural formulae cast from chemical analyses(Table denceof such Mn2+ bands are presentin the montmoril2) indicate that the three samplesare nearly pure mont- lonite spectra.It should be pointed out, however, that the
small Mn content (ca. 0. l0lo)would not show any absorpmorillonite. The layer chargesrangefrom -0.57 to -1.32
consistentwith the K-saturation results.Note that, in the tion bands if the Mn is in the 2+ oxidation state inasstructural formula representations,the number of octa- much as the Mn2+ ligand-field transitions are spin-forhedral cations are forced to total no more than 2.00 with bidden and have very low absorption coefficients.
The effect of Fe3+on the spectrashould also be menany extra Mg assignedto the interlayer.
Rnsur,rs AND DISCUSSToN
formulas
andstructural
compositions
TneLe2. Chemical
Diffuse-reflectancespectra of the montmorillonite
samplesNMNH 101836,NMNH F.7452,and K-P are Sample:
K-P.
R7452
101836
shown in Figures1,2, and 3, respectively.From the fit56.3
48.24
55.68
sio,
ting procedure(Table 3), absorption bandsare found near Al203
20.0
22.39
20.94
1.32
10400, 18800, and 20600, and 22400 cm-'. Several Fe.O"
0.50
0.42
samTreLe1. Observed
d spacingsin montmorillonite
ptes
Sample
060
N M N H1 0 1 8 3 6
NMNH R7452
1.490
1.490
HT
15.2
15.0
17
17
13.8
12.8
10.16
9.93
Note: AD : air dried; G : glycolsaturated;K : K saturated;
HT : heat treated at 550 "C for 2 h.
MnO
Mgo
CaO
Naro
KrO
H,O (tot.)
I otal
0.15
2.59
251
0.33
0.05
19.5
102.2
0.16
3.76
2.26
<0.1
0.25
22.4
100.3
1.38
2.63
0.75
0.69
0.52
16.4
100.0
Structuralformulas
NMNH 101 836: (Cao*Nao*Koo,)(Al3sMgo55MnoolrXSi?@Aloitoro(OH)4
NMNH R7452: (Cao€Mgo rxAls$Mgo65Mnod(Si? sAlo6')O4(OH)4'xHrO
K-P: (Cao,rMgoouNao,elG
lXAlssFeo,4MgosrMnooTXSiTeAlodoro(OH)4'xHrO
' Data for sample K-P, normalizedto 100%, are taken from Foord et al.
{1986).
142
SHERMANAND VERGO:Mn IN MONTMORILLONITE
W A V E L E N G T F I( N M )
500
oh
Dnt
Dnn
1000
o
K-P
KATFIINA MINE
PALA OIST..
CA
6
v
ul
o
z
o
o
a
o
_---
Erc\
-o'--
\
-drs
F
o)
tr
IJJ
20
15
LO
WAVENUMBEFIS,/
lOOO
25
Fig. 3. Diffuse-reflectance
spectraof sampleK-P (Foordet
al., 1986).Bandsin thespectra
areasin Fig. l.
tioned. The energiesof Fe3+ bands in montmorillonite
should be similar to those in nontronite. Octahedrally
coordinated Fe3+ in nontronite shows bands at 10600
cm-' (943nm), l6 100cm-' (621nm), 22500 cm-t (444
nm) corresponding
to the6A, - oTr,64r- oTz,and6AtoE, oAr ligand-field transitions (Sherman and Vergo, in
prep.; seealso,Karickhoffand Bailey, 1973;Singer,1982).
Note that the Fe3+ligand-field transitions, like those of
Mn2+, are nominally spin-forbidden. Bands due to small
amounts of Fe3+would be obscuredby the more intense
Mn3+ bands. Attempts to include bands due to Fe3+ligand-field transitions in the fitting procedurewere unsuccessful.
The Mn3+ cation has four 3delectrons;in an octahedral
field, the ground state 3d orbital electronic configuration
would be t)ret. Only a single spin-allowed electronic
transition would be possible, tE, - tTr* corresponding
to the lr, - econe-electronorbital transition. This should
occur near 20000..-t (500 nm) basedon SCF-Xa-SW
molecular orbital calculations (Sherman, 1984) and the
spectraof other Mn3+ minerals. As is well-known, however, the Mn3+ cation distorts its coordination site via
the Jahn-Teller effect.The distortion decreasesthe symmetry of the coordination site from octahedral (O) to
tetragonal (Do). Under the tetragonal distortion, the trlOo)
orbital splits into e,(Doo)and br,(Do) orbitals, whereas
€- o
b"n_
_
-
tF:
bzo
uo
Fig. 4. Mn3* 3denergy levelsin (a) tetragonaldistortion with
z ligands in, (b) regular octahedron,and (c) tetragonaldistortion
with zligands out. Orbitals that are unoccupiedare indicated by
dashedlines. The ligand-field theory parameter llDq (the t* e" orbital separationin 01 symmetry) is given by the br" - bru
orbital energyseparationin Doosymmetry.
the e,(O) orbital is split into a,"(Doo) and b,r(Doo) orbitals
(Fig. a). Hence, in a tetragonal site, three absorption bands
will be observed instead of one. The presence of four
bands in the montmorillonite spectra results from further
distortion of the tetragonal site to one with orthorhombic
(Cr,) or smaller symmetry. This further distortion splits
the etDo) orbital into singly degeneratea,(Cr) and b,(Cr)
orbitals.
From the band eneryies,it is possible to get an approximate estimate of the Mn-O bond lengths and the
degreeof distortion of the Mn3+ coordination site. To do
this, however, we must assign the bands to the specific
electronic transitions. For a first approximation, we shall
assumethat the octahedralMn3+ site has only undergone
the tetragonal distonion. Two different possible tetrago'
nal distortions of the MnO. coordination site must be
considered,however, depending upon whether the axial
Mn-O bonds are extendedor compressedrelative to the
equatorial Mn-O bonds. The one-electronenergy levels
for Mn3+ in both tetragonaldistortion schemesare shown
in Figure 4.
ln Duo(tetragonal)symmetry, the Mn 3d orbital energies are describedin terms of the parametersDs, Dt, and
Dq (Table 4), where D4 is the cubic crystal field splitting
and DJ, Dl describe the tetragonal field (Perumareddi,
Tneue3. Energies
andassignments
of Mn3*ligand-field
bands 1967;Konig and Kremer, 1977).If the tetragonaldistorin montmorillonite
spectra
tion is such that the axial Mn-O bonds are compressed
NMNH
NMNH
relative to the equatorial bonds (Fig. 4), then the bands
Assignment'
Don
uBrn' uAro
'Bro '
" B"o
58t 58, -
5Q
uBt '
.5E
5Ar
uA,
5Bt'5Bz
uA"
101836
R7452
10480
19041
20660
21837
10276
18751
20496
22127
K-P
10542
18560
2060s
22143
- The notationgivenis for the actual(multielectronic)
spectroscopicstates:
the Ble - A1s,Bs- B2sandBE - Eespectroscopictransitiongcorrespond
*
to the ale b,o,4o
b,o,and ee 4e one-electron-orbitaltransitions of
Mn3*in a tetragonalSite.
TaeLe4. Orbitalenergiesin the tetragonal
crystalfield
Orbital
4s
Energy
2Ds-Dt+6Dq
-2Ds-ODt+6Dq
2Ds-Dt-4Dq
-Ds+4Dt-4Dq
SHERMAN AND VERGO: MN IN MONTMORILLONITE
at 10400 and 22400 cm I correspondto the br, - ar,
and br" - drc transitions, and the two bands at 18 800
and 20600 cm-' correspondto the split €e- et, transition. Using the orbital energy expressionin Table 4, it
follows that llDq, Ds, and Dt are 12000, -1871, and
-583 cm I. The value for llDq in this energyJevel
scheme,however, is unrealistically small for a trivalent
cation octahedrally coordinated by oxygen.
If, instead, it is assumedthat the axial Mn-O bonds
are extendedrelative to the equatorial Mn-O bonds (Fig.
4c), the bandsat 10400 and 18800 cm-'would be assignedto rhe arr - br" and bt, - bt, transitions, whereas
the bands aI 20 600 and 22 400 cm-' would be split componentsof the e" - b," transition. This energyJevelscheme
implies that llDq, Ds, and Dt are 18 800, 1934,and 533
cm-1. The more reasonablevalue for llDq suggeststhat
this energy-levelschemeis correct.
The value for l}Dq for Mn3* in AlrO, is 19470 cm-'
(McClure, 1962),whereasl1Dq for the Mn3+ aquo-complex is 21 000 cm-' (Orgel, 1966).Hence, the value estimated for Mn3+ in montmorillonite (18800 cm-') is
quite reasonable.It should be pointed out, however, that
nearly all of the previous calculationsin the mineral literature of "l\Dq" for Mn3+ cations occupying distorted
sites in minerals have been done incorrectly. A common
mistake is to assume that llDq correspondsto the averageenergyof the er(Oo)-derivedstatesminus the average energy of the trr(Oo)-derivedstates.In the caseof a
tetragonallydistorted site, for example, this quantity calculated from the spectrawould not correspond to l}Dq
but instead would be l}Dq - 5Dt - Ds/2. For this reason, "10D4" for Mn3* cations are often quoted in the
mineral literature as being between 10000 and 15000
cm- ' instead of more realistic valuesbetween I 8 000 and
2 2 0 0 0c m - ' .
As noted previously, the presenceoffour bands shows
that the symmetry of the Mn3* coordination site is not
Do, but must be Cr" or lower. The assignmentsof the
bands in terms of Cr" symmetry is given in Table 3. The
Cr" symmetry, all the transitions are Laporte-allowed except the '8, - 'Br. The band at 20600 cm-l, therefore,
must correspondto that transition since it is much weaker than the other Mn3+ ligand-field bands.
We can now use the values of l}Dq and Dt to estimate
the degreeof tetragonal distortion in the Mn3+ site. The
tetragonal field parameter Dt is a measureof the differencebetweenthe Dqparametersassociatedwith the z-axis ligands and the t/-axes or equatorial ligands (e.g.,
Douglasand Hollingsworth, 1985):
Dt:
-%lDq(z) - Dq(xy)l : 533 cm-'.
If we assumethat the derived Dqparameter is an average
ofthat for the two oxygen types, then
Dq:
%l2Dq(z)+ 4Dq(xy)l: 1880cm-'.
Solving the two equations,we get Dq(xy): 2l9l cm-'
and Dq(z): 1258 cm-r. The significantdifferencebetween Dq(ry) and Dq(z) implies that the tetragonal dis-
t43
tortion of the site is quite large.Using the approximation
of ligand-field theory, we have Dq a (l/r)s where r is the
M-O bond length. Hence,
r(xy)/r(z) : lDq(z)/Dq(xy)l'd: 0.895.
The averageMn-O bond length should be about 2.04 L
given the ionic radius of Mn3* and the observed value
for l\Dq.It follows that r(xy) x 1.96 and r(z) x 2.19.
The ionic radius of Mn3* is nearly identical to that of
Fe3+.One might expect that a manganian (Mir3+) smectite (analogousto nontronite) might be stable or, at the
very least, that extensive solid solution between such a
hypothetical end member and nontronite could occur.
The absenceof such Mn3+ smectitesmay simply reflect
the limited stability of Mn3* in most geochemicalenvironments. In addition, however, the large tetragonal distortion of the Mn3* coordination site may limit the extent
of the solid solution of Mn3* in montmorillonite inasmuch as the z-axis elongation should strain the octahedral AI(O,OH) sheet.
Acxlqowr-BocMENTS
Thanks are due to P. Dunn (National Museum of Natural History) for
providing accessto samples 101836 and R7452 in the collection of the
National Museum E. Foord (U.S. Geological Survey) provided sample
K-P and associateddata for this material along with helpful discussions.
xnr analyseswere done by B. Scott (USGS)and total HrO determinations
were provided by Z A. Brown (USGS).S. Altaner, C' Lawson, M. Ross,
and K. Langer provided helpful comments.
RnrBnBNcnscrrpo
Brindley, G.W. (1980) Order-disorderin clay mineral structures'In G.W.
Brindley and G. Brown, Eds., Crystal structuresof clay minerals and
their X-ray identifrcation, p. 125-126. Mineralogical Society,London.
Bums, R.G. (1970) Mineralogical applications of crystal field theory'
CambridgeUniversity Press,Cambridge.
Douglas,8.F., and Hollingsworth, C.A. (1985) Svmmetry in bonding and
spectra,an introduction. Academic Press,Orlando, Florida.
Foord, E.E., Starkey,H C., and Taggart,J.E. (1986) Mineralogy and paragenesisof "pocket" claysand associatedminerals in complex granitic
pegmatites,San Diego County, California' American Mineralogist, 71,
428-439.
HAlenius,U. (1978) A spectroscopicinvestigation ofmanganian andalusite. CanadianMineralogist, 16, 567-57 5.
Jackson, M.L. (1956) Soil chemical analysis-Advanced course. Published by the author, Department of Soil Science,University of Wisconsin, Madison, Wisconsin.
Karickhoff, S.W., and Bailey, G.W. (1973) Optical absorption spectraof
clay minerals. Clays and Clay Minerals, 21,59-70.
Keester,A.S, and White, W.B. (1968) Crystal field spectraand chemical
bonding in manganeseminerals.In Papersand Proceedingsofthe Fifth
GeneralMeeting, International MineralogicalAssociation, 1966,p. 2235. Mineralogical Society,London.
Kinter, E.B., and Diamond, S. (1956) A new method for preparationand
samplesof soil clays for X-ray diffractreatment of oriented ag1regate
tion analysis.Soil Science,81, I I 1-120.
Konig, E., and Kremer, S. (1977) Lignd field energydiagrams' Plenum
Press,New York.
McClure, D.S. (1962) Optical spectraof transition metal ions in corundum. JournalofChemical Physics,36, 2757-2779.
Orgel, L.E. (1966) An introduction to transition metal chemistry: Ligand
field theory. Methuen, London.
Perumareddi,J.R. (1967) Ligand field theory ofd3 and d electronicconfigurationsin noncubic fields: L Wave functions and energymatnces.
Journal of PhysicalChemistry, 7 l, 3144-3154.
t44
SHERMAN AND VERGO: MN IN MONTMORILLONITE
Ross, C.S., and Hendricks, S.B. (1945) Mineials of the montmorillonite
group and their origin and relations to soils and clays.U.S. Geological
Survey ProfessionalPaper 205-8,79 p.
Sherman,D.M. (1984)Electronicstrucluresof manganeseoxide minerals
American Mineralogist, 6'l , 788-799.
Singer,R.B. (1982) Spectralevidencefor the mineralogy ofrhe high albedo dusts and soils on Mars Joumal of GeophysicalResearch,87,
10159-10168.
Weaver, C.E., and Pollard, L.D. (1973) The chemistry of clay minerals.
Developmentsin Sedimentology,15,213 p.
M,c.NuscRrpr REcETVEDOcrosrn
MeNuscnrsr
22, 1986
23, 1987
AccEprED SsprEMsrn